• The X11 server runs on your local machine and accepts display updates from the remote server, while passing your inputs to the remote server.
  • X11 forwarding over SSH lets you execute a command on a remote server while creating a secure tunnel for the graphical output.
• Many HPC applications have GUIs that require X11 to display on your local machine.
  • Matlab Desktop, Mathematica Notebooks, gnuplot, etc.

If you have a Windows computer, there are many applications that can enable X11 Forwarding

• Windows Subsystem for Linux: A Windows native compatibility layer for running Linux binaries on Windows.

  • My recommendation ❤️

• MobaXterm: MobaXterm is a free, enhanced terminal for Windows with X11 server, SSH client, network tools, and much more.

  • 😎 It’s a popular choice among HPC users

• Xming: Xming is X Window System Server for Microsoft.

• Cygwin: Cygwin is a free Linux-like environment for Windows. To add Cygwin/X server, select the xinit package from the X11 category.

• Install
  • Start a Windows Command Prompt in administrator mode
  • Execute wsl --install
• Operation
  • Open the WSL2 application
  • This is essentially a Linux kernel functioning within your Windows machine
  • Login to Hoffman2 with ‘-Y’
    • Type ssh -Y joebruin@hoffman2.idre.ucla.edu

For Mac users, XQuartz is a popular application that supports X windows.

This is a terminal application that facilitates a connection to Hoffman2 with X11 Forwarding enabled.

Once installed, to connect to Hoffman2 and enable X11 Forwarding, run ssh with the -Y option:

ssh -Y joebruin@hoffman2.idre.ucla.edu

NoMachine provides remote desktop functionality on Hoffman2, enabling a Graphical User Interface (GUI) on a remote machine.

• Download NoMachine
  • Use the download link for your operating system (Windows, Mac OSX, or Linux)
• After downloading, install and configure the NoMachine client to connect to Hoffman2:
  • Click on Add
  • Set up a Name for your connection
  • The Host should be nx.hoffman2.idre.ucla.edu
  • Set the Port to 22 and choose SSH for Protocol

• After creating this configuration, click on the new icon in the list of machines:
  • You may be prompted to Verify host identification
    • If so, click OK
  • Enter your Hoffman2 Username and Password
  • You’ll then be asked to Create a new desktop
    • Choose to Create a new Mate virtual desktop
  • Upon completion, you’ll be presented with a desktop view of your Hoffman2 account:
    • Open a terminal to start any GUI applications
      • However, you may want to initiate an interactive job first
        • To do this, type qrsh -l h_data=10G

Jupyter can be executed on Hoffman2. This process involves running the jupyter notebook/lab command on a Hoffman2 node, forwarding the Jupyter port to your local machine, and accessing Jupyter via your local web browser.

However, for a more streamlined experience, we have created a script, h2jupynb, which automatically creates a Jupyter session on a Hoffman2 compute node.

wget https://raw.githubusercontent.com/rdauria/jupyter-notebook/main/h2jupynb
chmod u+x h2jupynb

Ensure Python is installed on your local machine before executing the script:

python h2jupynb -u joebruin -t 5 -m 10 -a "intel-gold\*"

The above command will initiate the script with your Hoffman2 username (replace “joebruin” with your username), a duration of 5 hours (-t 5), memory request of 10 GB (-m 10), and specify the architecture as “intel-gold”.

Usage:

h2jupynb [-u <Hoffman2 user name>] [-v <python-version>]
         [-t <time, integer number of hours>] [-m <memory, integer number of GB per core>]
         [-e <parallel environment: 1 for shared, 2 for distributed>] [-s <number of slots>]
         [-o <run on group owned nodes: yes/no>] [-x <run on an exclusively reserved node: yes/no>]
         [-a <CPU-type>] [-d <path to directory from which the jupyter NB/lab will start>]
         [-g <run on a gpu node: yes/no>] [-c <gpu-card-type>] [-l <cuda-version>]
         [-p <port number>]

To use your Anaconda environment created on Hoffman2 with the h2jupynb script, you’ll need to install ipykernel in your Anaconda environment.

Let’s say your Anaconda environment is named myconda. Follow the steps below on Hoffman2:

Request an interactive session with 10GB of memory for 1 hour:

qrsh -l h_data=10G,h_rt=1:00:00

Load Anaconda and activate your environment:

module load anaconda3
conda activate myconda  

Install ipykernel and add it to your environment:

conda install ipykernel
python -m ipykernel install --user --name=myconda --display-name "myconda"

Run the h2jupynb script on your local machine using the -v anaconda3 option:

python h2jupynb -u joebruin -v anaconda3

Now, myconda should appear in the list of available kernels in Jupyter.

For more information on using Jupyter on Hoffman2, check out our Hoffman2 documentation

You can conveniently interact with Hoffman2 using RStudio Server, allowing you to utilize R in a familiar, intuitive environment.

Find detailed instructions for using RStudio on Hoffman2 on our GitHub page

We’ve also conducted a dedicated workshop on using Rstudio on Hoffman2. Feel free to explore this resource for additional insights and usage tips.

Paraview is an open-source application that supports interactive, scientific visualization.

For using Paraview, you will first need to install ParaView on your local machine.

We will run Paraview on a compute node as a server process. Then use our local install of Paraview to connect to Hoffman2.

Initiate an interactive job on a compute node and load the ParaView module.

qrsh -l h_data=10G
echo $HOSTNAME #Take note of the hostname of the compute node you landed on
module load apptainer
apptainer exec $H2_CONTAINER_LOC/h2-paraview_5.10.0-cpu.sif pvserver --server-port=11111 

Leave this terminal open.

Open another local terminal for port forwarding:

ssh -L 11111:nXXXX:11111 joebruin@hoffman2.idre.ucla.edu

Replace nXXXX with the hostname of the compute node and joebruin with your Hoffman2 username.

• Then you will start up ParaView on your local machine
  • This MUST be version 5.10.0 of paraview (though 5.10.1 may work).
• Click on Connect
  • Create a Server with localhost:11111

To utilize the GPU compute nodes, follow the steps below:

qrsh -l h_data=5G,gpu,V100
apptainer exec --nv $H2_CONTAINER_LOC/h2-paraview_5.10.0-gpu.sif pvserver --server-port=11111

Follow the same port forwarding and connection steps as described above.